🤷

Original by GihanZ on Reddit.

This week’s AI news:

🔸Google’s toolkit to build autonomous agents
🔸Newfound risks in AI-generated code
🔸All about RA-DIT, a new fine-tuning method

Plus: Andrew Ng discusses multi-agent collaboration, a key AI agentic design pattern.

Read #TheBatch : hubs.la/Q02tf_Pw0

In Quantization Fundamentals with @HuggingFace, you'll learn how to quantize open source models, to make them more accessible and efficient.

You’ll get hands-on and practice by quantizing open source multimodal and language models.

Start now! ➡️ hubs.la/Q02t8Mxd0

Emad Mostaque resigned as CEO of Stability AI, the creator of Stable Diffusion, amidst revenue shortfalls, decreasing confidence from investors, and departure of key personnel.

Find out more details in #TheBatch : hubs.la/Q02t48690

LLMs can take gigabytes of memory to store, which limits what can be run on consumer hardware. But quantization can dramatically compress models, making a wider selection of models available to developers. You can often reduce model size by 4x or more while maintaining reasonable…

Researchers from Carnegie Mellon University and Princeton University introduced the Mamba architecture, an approach that challenges traditional transformers in processing efficiency and memory usage.

Read our summary of the paper in #TheBatch ⬇️ hubs.la/Q02sWTCB0

A study offered insights into how users are currently adopting generative AI.

While text generation remains dominant, users are surprisingly slightly more inclined to generate videos than images.

Find more highlights from the study in #TheBatch : hubs.la/Q02sLDbm0

Weekend course alert: 'Knowledge Graphs for RAG' by DeepLearning.AI and #Neo4j

The course will help you use #KnowledgeGraphs to uncover deeper insights into your data and enhance the performance of #LLMs with structured, relevant context.
bit.ly/3JdRQ2Z

#RAG #GenAI

A new generation of coding tools is emerging, harnessing the capabilities of LLMs to automate complex software programming tasks through agent-style interactions.

Learn more in #TheBatch : hubs.la/Q02sDyLS0

ICYMI, yesterday we launched a short course with UnstructuredIO.

Learn techniques for representing all sorts of unstructured data, like text, images, and tables, from many different sources and implement them to extend your LLM RAG pipeline.

Enroll now: hubs.la/Q02sC7Sg0

Read Data Points to find the latest AI updates of the week, including:

👉 Advanced new editing tools by DALL·E
👉 Command R+, a new LLM by Cohere
👉 An update on big tech's hunt for AI training

And more!

Check out Data Points now: hubs.la/Q02sz2510

This week's AI news:
🧑‍💼 Instability at Stability AI
💻 AI-powered coding tools enhance software development
🤖 A new approach to outperform transformers

Plus: Andrew Ng discusses planning, a key design pattern seen in AI agents.

Read #TheBatch : hubs.la/Q02stZ0s0

Real-world data comes in many formats, such as PDF, HTML, Powerpoint, markdown, etc. Creating preprocessing pipelines for each of these file formats to enrich your RAG system can result in widely different content structures, unaligned chunking strategies, and unnecessary costs.…

Join 'Preprocessing Unstructured Data for LLM Applications' today: hubs.la/Q02skXyY0

Introducing Preprocessing Unstructured Data for LLM Applications, a short course in collaboration with UnstructuredIO!

Learn techniques for handling unstructured data, including images, tables, and documents like PDF and PowerPoint.

Join today: hubs.la/Q02skXyY0

Researchers at Stanford University and UNC-Chapel Hill developed a new fine-tuning method that makes LLMs more likely to stick to the facts. Instead of RLHF, FactTune uses a combination of DPO and RLAIF, making fine-tuning less costly and more efficient.

Read more: hubs.la/Q02sjxms0

When you join our courses, you're also becoming part of an active, supportive community.💡

Recently, a learner posed a question about our course with @Predibase, and instructor Travis Addair jumped in with an answer himself.

Join our community 👉 hubs.la/Q02sgrl-0

In 'Pair Programming with a Large Language Model,' instructed by Google’s Laurence Moroney, you will learn how to use LLMs to simplify your code and reduce technical debt by explaining and documenting a complex existing code base.

Sign up today: hubs.ly/Q02s9CpK0

In response to concerns about the potential misuse of AI in producing biological materials, synthetic biologists from Asia, Europe, and North America signed a pledge to regulate their use of AI in designing proteins.

Find more details in #TheBatch : hubs.la/Q02s7Wyz0